https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126123
--- Comment #5 from Avinash Jayakar <avinashd at gcc dot gnu.org> --- (In reply to Iain Sandoe from comment #4) > Is there ever a case in the GCC code where we will emit EH = 1? Yes, with the change in r17-1992-ge4543a6ecc97a75 we introduce the new ppc specific builtin "__builtin_ppc_atomic_cas_local", where EH bit for l<>arx will be 1. But in all other cases it would be 0, and there are no optimizations in gcc, that would change the EH bit to 1 as of now.
